Epidemiol. Infect. (1990). 105. 139-162 139 Printed in Great Britain Health effects of beach water pollution in Hong Kong W. H. S. CHEUNG, K. C. K. CHANG, R. P. S. HUNG Environmental Protection Department, 28/F., Southorn Centre, Wanchai, Hong Kong AND J. W. L. KLEEVENS* Department of Community Medicine, University of Hong Kong (Accepted 24 December 1989) SUMMARY Prospective epidemiological studies of beach water pollution were conducted in Hong Kong in the summers of 1986 and 1987. For the main study in 1987, a total of 18741 usable responses were obtained from beachgoers on nine beaches at weekends. The study indicated the overall perceived symptom rates for gastrointestinal, ear, eye, skin, respiratory, fever and total illness were significantly higher for swimmers than non-swimmers; and the swimming- associated symptom rates for gastrointestinal, skin, respiratory and total illness were higher at 'barely acceptable' beaches than at 'relatively unpolluted' ones. Escherichia coli was found to be the best indicator of the health effects associated with swimming in the beaches of Hong Kong. It showed the highest correlation with combined swimming-associated gastroenteritis and skin symptom rates when compared with other microbial indicators. A linear relationship between E. coli and the combined symptom rates was established. Staphylococci were correlated with ear, respiratory and total illness, but could not be used for predicting swimming- associated health risks. They should be used to complement E. coli. The setting of health-related bathing-water quality standards based on such a study is discussed. INTRODUCTION Swimming is Hong Kong's most popular summer recreation. There are 42 coastal beaches gazetted for such a purpose. No fewer than 18 million person-visits are being recorded at these beaches each year. Popular beaches may receive up to 1-5-3 million visitors in a season, and 50000 or more on a single day. In connection with protecting the public from waterborne infections, the most important beach- water quality standards are microbiological in nature. To enable these standards to be met, expensive sewage treatment and disposal facilities near bathing beaches have been built, and sewerage schemes at a total cost of around HK$700 million are being planned. First adopted in 1981, the existing bacterial water quality objective for the * Present address: Department of Epidemiology and Social Medicine, University of Antwerp, Belgium. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737 Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Transcript
Epidemiol. Infect. (1990). 105. 139-162 1 3 9Printed in Great Britain
Health effects of beach water pollution in Hong Kong
W. H. S. CHEUNG, K. C. K. CHANG, R. P. S. HUNG
Environmental Protection Department, 28/F., Southorn Centre, Wanchai,Hong Kong
AND J. W. L. KLEEVENS*
Department of Community Medicine, University of Hong Kong
(Accepted 24 December 1989)
SUMMARY
Prospective epidemiological studies of beach water pollution were conducted inHong Kong in the summers of 1986 and 1987. For the main study in 1987, a totalof 18741 usable responses were obtained from beachgoers on nine beaches atweekends. The study indicated the overall perceived symptom rates forgastrointestinal, ear, eye, skin, respiratory, fever and total illness weresignificantly higher for swimmers than non-swimmers; and the swimming-associated symptom rates for gastrointestinal, skin, respiratory and total illnesswere higher at 'barely acceptable' beaches than at 'relatively unpolluted' ones.Escherichia coli was found to be the best indicator of the health effects associatedwith swimming in the beaches of Hong Kong. It showed the highest correlationwith combined swimming-associated gastroenteritis and skin symptom rates whencompared with other microbial indicators. A linear relationship between E. coli andthe combined symptom rates was established. Staphylococci were correlated withear, respiratory and total illness, but could not be used for predicting swimming-associated health risks. They should be used to complement E. coli. The setting ofhealth-related bathing-water quality standards based on such a study is discussed.
INTRODUCTION
Swimming is Hong Kong's most popular summer recreation. There are 42coastal beaches gazetted for such a purpose. No fewer than 18 million person-visitsare being recorded at these beaches each year. Popular beaches may receive up to1-5-3 million visitors in a season, and 50000 or more on a single day. In connectionwith protecting the public from waterborne infections, the most important beach-water quality standards are microbiological in nature. To enable these standardsto be met, expensive sewage treatment and disposal facilities near bathing beacheshave been built, and sewerage schemes at a total cost of around HK$700 millionare being planned.
First adopted in 1981, the existing bacterial water quality objective for the
* Present address: Department of Epidemiology and Social Medicine, University of Antwerp,Belgium.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
bathing beaches in Hong Kong is as follows: ' The level of Escherichia coli shouldnot exceed 1000 per 100 ml, calculated as the running median of the most recent5 consecutive samples. Samples should be taken 3 times in any one month atintervals of between 3 and 14 days during the bathing season' [1]. This standardwas set by modifying the ' interim' acceptability criterion recommended by aWHO/UNEP report [2], at a time when epidemiological information on the healtheffects of beach water pollution was very sparse indeed.
Subsequently, an annual acceptability standard of 60% compliance with the1000 E. coli per 100 ml criterion was also adopted in Hong Kong. Only thosebeaches which could not attain this minimum standard were ' closed' by the twomunicipal councils, the management authorities for gazetted bathing beaches.Warning signs would be erected at these beaches, and life-guards as well as showerfacilities would be withdrawn, so as to discourage the bathers from entering intothe water.
The existing Hong Kong beach-water quality standards, as well as the WHOand EEC standards, were set based on judgement and were not supported by dataon health effects due to swimming exposure. The WHO and UNEP reports [2,3]putting forward the ' interim' acceptability criterion recommend epidemiologicalstudies of beach water pollution be undertaken by individual countries, so thathealth-related bathing-water criteria could be developed.
There are two types of epidemiological investigations related to beach waterpollution, retrospective and prospective studies. In a retrospective study,individuals who have developed a disease are compared with a group of similarindividuals who did not, with respect to exposure to bathing waters. Theusefulness of such type of studies for developing health-related beach-watercriteria is limited, as they are unlikely to yield dose-response relationships [2]. Itis also not possible to use this method for studying every possible healthconsequence of swimming exposure, and one would need to identify the diseasesthat are to be investigated at the outset of the study. These were poliomyelitis andenteric fever in the well known study by the Public Health Laboratory Service(PHLS) [4].
On the other hand, retrospective analysis of recreational waterborne outbreaksis important for revealing the specific disease entities which could be transmittedthrough swimming in faecally polluted waters. For instance, there were studiesindicating that enterovirus-like illnesses (particularly those caused by coxsackie-viruses) were swimming-related [5-7]. Other studies also provided evidence thatswimming in polluted freshwater beaches could be linked with outbreaks ofhepatitis A [8], shigellosis [9,10], and Norwalk virus gastroenteritis [11,12].
In a prospective study, groups differing in their exposure to bathing water arefollowed up, and illness incidence is compared in relation to such exposure. Therewere two such studies based on which health-related beach-water qualitystandards were set. The first one was undertaken by the US Public Health Serviceduring 1948-50, in which illness information from 22264 participants wasobtained [13]. It was found at a freshwater Chicago beach that symptomatic ratesfor ear, eye, skin, respiratory and gastrointestinal illnesses were significantly higherin individuals who swam on 3 successive days when the geometric mean 'totalcoliform' density was 2300 per 100 ml, than in swimmers who swam on 3 successivedays when the mean coliform density was 43 per 100 ml. In the Ohio River study.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Beach water pollution in Hong Kong 141a detectable health effect was found amongst river swimmers; they displayed 32 %more gastrointestinal illness than might be expected by chance. The median totalcoliform density in the stretch of the river was 2700 per 100 ml. No relationshipsbetween total coliform levels and swimming-related illness were observed atmarine beaches. Based on these findings, the then Federal Water PollutionControl Administration arrived at a standard of 200 'faecal coliforms' per 100 mlfor both marine and freshwater beaches in 1968 [14]. This criterion was heavilycriticized - in respect of the inadequacy in the design of the original studies; thepaucity of valid epidemiological data; the ways in which the criterion was derived;and the microbial indicator used [4,15,16].
The second study was conducted by the workers of US EPA during 1973-82,with a view to correct the deficiencies of the previous survey. The study, this timeon marine beaches, involved obtaining usable responses from 26686 beachgoers atfive beaches [16,17]. It was found gastrointestinal symptoms were both swimming-associated and pollution-related. Enterococcus was the best indicator as it showedthe highest correlation with swimming-associated gastroenteritis symptom ratesand a linear relationship was established. The US workers also concluded theprevious use of the criterion of a geometric mean faecal coliforms density of 200per 100 ml would have caused an estimated 19 cases of acute gastroenteritisdisease per 1000 swimmers at marine beaches, and the US community has' unknowingly accepted' this swimming-associated health risk. The geometricmean enterococci density corresponding to this historically accepted risk is 35 per100 ml; and this is the new standard for marine beaches recently recommended byUS EPA in 1986 [18], 18 years after the previous standard was set. Similar studieswere conducted at four marine beaches in Alexandria, Egypt (which are generallymore polluted than the US beaches), in which usable responses were obtained from23080 beach visitors [16]. Strong association of diarrhoea and vomiting symptomsto E. coli as well as enterococci densities was observed. Another study wasundertaken at four US freshwater beaches with 38140 usable responses obtained[19,20]. There was strong correlation between 'highly credible' gastrointestinalsymptoms and E. coli as well as enterococci; and both E. coli and enterococcistandards have been recommended for the freshwater beaches of USA [18].
Subsequent prospective studies of similar nature undertaken by workers inCanada [21,22], Israel [23] and England [24] were of smaller scale when comparedto the US EPA studies; and their findings have not been used for setting beach-water quality standards. The number of beachgoers interviewed at the freshwaterbeaches in Canada and the coastal beaches in Israel and England were 4537, 2036and 1903, respectively.
This paper reports the findings of the prospective epidemiological studies ofbeach water pollution undertaken in Hong Kong in 1986 and 1987. It firstlyascertains whether swimming in the subtropical beach waters of Hong Kongcarries an increased risk of illness; and whether swimming in faecally pollutedbeaches poses higher health risks than in less polluted ones. Secondly, itdetermines which microorganism is the best indicator of the health effects of beachwater pollution; and whether a quantitative relationship between the densities ofthis indicator organism in beach waters and swimming-associated health risks canbe established. Finally, the use of such a relationship for setting health-relatedbathing-water quality standards in Hong Kong is discussed.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
The method for the study was based on the protocol recommended byWHO/UNEP reports [2,25], with some modifications to suit local conditions.
The study was conducted in two phases. The initial phase (pretest) wasundertaken at four very popular beaches, during six weekends in July and Augustof 1986. The purpose of this phase was to test the epidemiological techniques, andto provide preliminary estimates of the swimmer and non-swimmer (background)illness rates needed for estimating the sample size for the main study. The secondphase study was carried out at nine popular beaches, over 11 weekends from Juneto September of 1987. This main study in 1987 was to provide the data for settinghealth-related bathing-water quality standards which suit the local conditions ofHong Kong.
For each phase, the studies undertaken were divided into two parts:epidemiological studies involving beach and follow-up telephone interviews; andintensive monitoring of the microbiological quality of beach waters over weekends.
Beaches under studyThe pretest in 1986 was carried out at Repulse Bay, Stanley Main, Lido and Old
Cafeteria Beaches. The nine beaches in the major study in 1987 were Repulse Bay,Deep Water Bay, Stanley Main, Shek O, Clear Water Bay, Lido, Butterfly, andOld and New Cafeteria. The location of these beaches is shown in Fig. 1. Thesebeaches, with their levels of microbial pollution falling on a gradient, were selectedbased on the bacterial water quality data obtained over the years by theEnvironmental Protection Department of the Hong Kong Government.
Repulse Bay, Deep Water Bay, Stanley Main, Shek 0, Clear Water Bay andLido Beaches are polluted to different extent by human sewage discharged fromsubmarine sewage outfalls nearby, or carried by stormwater drains running intothe beaches. Old and New Cafeteria Beaches are contaminated by livestock wastes(mainly pig excreta) discharged from Tuen Mun Nullah. Butterfly Beach isaffected by both Pillar Point submarine sewage outfall and Tuen Mun Nullah.
Water samplingSubsurface water samples were collected from three sampling points (50-150 m
apart, 1 m deep) at each beach under study, in locations with high densities ofbathers. For each beach, this was carried out every 2 h from 9 am to 5 pm. onweekend days in which beach interviews were carried out. The samples werepacked on ice, kept in the dark, and processed within 4-6 h after collection.
Microbiological analysisA total of nine microbial indicators in beach water samples were analysed,
namely faecal (thermotolerant) coliforms, E. coli, Klebsiella spp., faecal strepto-cocci, enterococci, staphylococci, Pseudomonas aeruginosa, Candida albicans andtotal fungi. All these indicator organisms in beach water samples were enumeratedusing the membrane filtration method. The 0-75 /iva Millipore HC membranefilters were used for testing faecal coliforms, E. coli and Klebsiella spp.; and the restof the microorganisms were analysed by the 0-45 /im Gelman GN-6 membranefilters.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Fig. 1. Location of the nine bathing beaches under study in Hong Kong.# , Beach; ->, submarine outfall.
The faecal coliforms were counted as the yellow colonies on the mLS medium[26,27], after incubation at 44 °C for 18 h. The colonies on the same membraneremaining yellow after the in situ urease test were enumerated as E. coli [28, 29].The Klebsiella spp. were those yellow thermotolerant colonies which turned pinkin the in situ urease test.
Faecal streptococci were analysed using the KF medium [30]; and forenterococci, the m-Enterococcus medium [31]. The VJP medium [32,33] was usedfor enumerating staphylococci; the mPA-E medium [34,35], for Pseudomonasaeruginosa; the mCA medium [36], for Candida albicans; and the MSTMEAmedium [37], for total fungi.
Selected microbial colonies on the membranes had been confirmed by the APIor VITEK identification systems.
The interviewsThe beachgoers were first recruited at the beaches during weekends. Their
names, addresses and telephone numbers were obtained. They were also askedabout possible pre-trial illness, and pre-trial swimming activities in the weekbeforehand.
Follow-up telephone interviews were conducted the next day for solicitinginformation on water exposure while at the beach during the weekend. Therespondents were asked when they had swum; and whether they had swum withhead above water, or just wading without getting their head into the water. Thiswas important in classifying the beachgoers as 'swimmers' or 'non-swimmers'.Further information on their demographic variables and on the type of food eatenat the beach was also solicited.
The beachgoers were again telephone interviewed 7-10 days afterwards, to see
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
whether they had any illness developed, and to obtain information on the on-settime of the perceived symptoms and their duration. Possible post-trial mid-weekswimming activities were asked about.
Definition of ' swimmers' and ' non-swimmers'Swimming has been defined as having significant exposure of upper body orifices
(i.e. the head) to beach water, in accordance with the recommendation ofWHO/UNEP reports [2,25]. 'Swimmers' therefore included individuals whoentered the water and immersed their heads on the day of study, and also youngchildren who did not actually swim in the strict sense of the word, but who spenttime in the water or at the water's edge and whose faces were splashed by waveaction or other means, thus affording a chance to ingest beach water. 'Xon-swimmers' were those beachgoers who had not immersed their heads in beachwater.
Symptoms and swimming-associated illness ratesThe perceived symptoms reported were grouped into the following categories:
gastrointestinal (GI), highly credible gastrointestinal (HCGI), ear. eye. skin,respiratory, and others (mainly fever and headache).
Gastrointestinal (GI) symptoms include vomiting, diarrhoea, stomach ache andnausea. 'Highly credible gastrointestinal' (HCGI) symptoms are denned as anyone of the following: vomiting; diarrhoea with a fever or a disabling condition(remained home, remained in bed or sought medical advice); and nausea orstomach ache accompanied by a fever. Individuals in this perceived symptomcategory are considered to have acute gastroenteritis.
Ear symptoms include earache or discharge; and for eye symptoms, eye-sore,irritation or redness. Skin symptoms are mainly rashes, exclusive of sunburn.Respiratory symptoms cover sore throat, heavy cough, and running nose.
The swimming-associated illness rate (or swimming-related health risk) for eachsymptom category was obtained by subtracting the perceived illness rate for non-swimmers from that for swimmers. The relative risk (RR) was the ratio of theillness rates for swimmers to those for non-swimmers.
Statistical analysisStatistical significance was analysed using the t test. STATGRAPHICS software
was used for constructing deterministic regression models and testing theirsignificance.
RESULTS
Pretest, 1986
A total of 6639 beachgoers were interviewed at the four beaches, and 5114follow-up interviews were completed. The response rate was 77 %. Around 30% ofthe beachgoing population swam in the mid-week before or after the weekendtrials, and hence had multiple exposure to various beach or swimming-pool waterswith varying microbiological quality. Their responses were excluded from thestudy. Amongst the 3549 usable responses, 73 % were from swimmers and the restfrom non-swimmers. The perceived symptom rates for swimmers were higher than
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Beach water pollution in Hong Kong 145those for non-swimmers in the categories of GI, HCGI, eye, skin, fever and totalillness. The swimming-associated gastroenteritis (HCGI) rates for various beacheswere low, ranging from 0-5 to 3-1 per 1000 swimmers; and the overall rate was 1-3per 1000 swimmers. An analysis of the pretest data indicated if significantdifference in the symptom rates between swimmers and non-swimmers were to beobtained in the main study, at least 32800 beachgoers should be interviewed and17 700 usable responses obtained (type-one error, a = 5%; type-two error, /? =10%).
Second phase study, 1987Populations sampled
A total of 33083 beachgoers were preliminarily interviewed at the nine beachesin this phase, and 24308 follow-up interviews were completed. The overallresponse rate was 73%. After excluding those respondents who swam in the mid-week before or after the weekend trials, the number of usable responses was18741, representing 77 % of the total. The swimmer and non-swimmer numbers atthe nine beaches were 14464 (77%) and 4277 (23%), respectively. This ratio ofswimmers to non-swimmers at the subtropical beaches of Hong Kong was verydifferent from those observed in the studies at the temperate coastal beaches ofEngland [24] and freshwater beaches of Canada [21], where the swimmers onlyrepresent 13% and 43% the beachgoer populations, respectively.
Swimming habits and demographic informationIt was found the beachgoers in Hong Kong on average spent 3'5 h on the nine
popular beaches; and the 'swimmers' usually swam in beach water for 1-3 h (theduration of exposure) at weekends.
Amongst the 24308 beachgoers interviewed, a total of 18392 (76%) belonged tofamily groups. The number of families participating in the study was 5284.
The non-swimming controls differed from swimmers in the distribution of sex;occupation; and the percentage eating barbecue food at the beaches. These factorswere however not important in affecting the swimming-associated symptom rates,as the overall rates unadjusted and adjusted for demographic variables were notsignificantly different.
Microbial beach-water qualityThe overall geometric mean densities of the nine microbial indicators at the
beaches under study are given in Table 1. The beaches were listed in the order ofincreasing levels of pollution as indicated by their mean E. coli counts. Othermicrobial indicators (e.g. faecal streptococci and staphylococci) however gavedifferent pictures on the relative degrees of pollution of these beaches. Forinstance, the mean staphylococci levels of Lido and Clear Water Bay Beaches werehigher than those of Old Cafeteria Beach, even though the latter was mostseriously polluted in terms of E. coli densities.
Relationships between microbial indicatorsThe relationships between various microbial indicators in beach waters were
assessed (see Table 2). There was high correlation amongst the commonlyrecognized indicators of faecal pollution, namely faecal coliforms, E. coli, faecalstreptococci and enterococci; but low amongst the various microbial indicators
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
which may come from both faecal and non-faecal sources, such as staphylococci.Pseudomonas aeruginosa, total fungi and Candida albicans. The correlationbetween these two categories of indicator organisms (for instance, between E. coliand staphylococci) was low.
An analysis of the E. coli to faecal coliforms ratios for the 667 samples revealed,on average, E. coli only represented 57% of the faecal coliform group in thebeach waters of Hong Kong. The significance of this low percentage in subtropicalwaters is discussed in a separate paper [29]. The mean enterococci to faecalstreptococci ratio for these beach-water samples was 144%, indicating the countsobtained using the KF medium were generally lower than those obtained with them-Enterococcus medium in analysing Hong Kong beach waters.
Symptomatic illness ratesThe perceived symptom rates obtained for the nine beaches are presented in
Table 3. The rates for most of the symptom groups were generally higher forswimmers than non-swimmers. For individual symptom categories, the overallperceived illness rates for swimmers were significantly higher than those for thenon-swimming control group. The HCGI symptom rate for swimmers was 5 timeshigher than that for non-swimmers; for eye or fever symptoms rate, 4 times: andfor GI, skin, respiratory or total illness rate, 2-3 times. There was also a significantexcess of total illness for swimmers than non-swimmers at each of the ninebeaches.
Apart from Old Cafeteria Beach, which is mainly polluted by livestock wastes,significant swimming-associated GI symptom rates were observed for the morepolluted beaches (Lido, Repulse Bay and Butterfly), as indicated by their overallgeometric mean E. coli counts.
The beachgoers did not report on symptoms suggestive of serious disease suchas shigellosis or enteric fever.
Onset and duration of symptomsThe incubation periods and duration of the perceived symptoms amongst the
beachgoers are presented in Table 4. The HCGI symptoms (or gastroenteritis)developed among swimmers were characterized by their short incubation periods,with a median of only 2 days. The duration of the HCGI symptoms was around2 days. As for the skin rash symptoms, they were usually developed after 1 day,and lasted for 4 days amongst the bathers.
Age distribution of symptomsThe swimming-associated symptom rates for different age groups are presented
in Fig. 2. The swimming-associated GI, HCGI, skin, respiratory, fever and totalillness symptoms were more prevalent among children under 10 years of age thanolder bathers (see Table 5).
Risks at ' relatively unpolluted' and ' barely acceptable' beachesTo see whether swimming in more polluted waters poses higher health risks than
in less polluted ones, the beaches under study were grouped into two categories,'relatively unpolluted' (RU) and 'barely acceptable' (BA). This grouping wasundertaken for each of the microbial indicators. The threshold level of an indicator
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Table 4. Incubation periods and duration of perceived symptoms in days,Hong Kong study 1987
SymptomsGastrointestinal
HCGIOverall
Ear and eyeEarEye
Skin rashRespiratoryFever
f
Numberreporting
2378
1366
176228116
SwimmerA
Medianincub-ationperiod
21
21122
Averageduration
2-31-9
1-52-9403-54-2
Non-swimmerA
{
Numberreporting
26
05
262510
Medianincub-ationperiod
4< 1
—1122
Averageduration
4-53-2
—3-25-24-23-8
Skin Resp. TotalFig. 2. Swimming-associated symptom rates for different age groups at nine Hong
Kong beaches, 1987. •> 0-4 years; Q, 5-10; 03, 11-25; 13, > 25.
for differentiating RU from BA beaches is the lowest density with the greatestnumber of symptom groups showing significant difference in the swimming-associated illness rates for the two categories of beaches (see Table 6 for thedetermination of the thresholds for E. coli and staphylococci). It is the microbialdensity beyond which the risks of swimming-related diseases are expected toincrease significantly. A geometric mean E. coli density of 180 per 100 ml wasfound to be the threshold for differentiating BA from RU beaches, with the formershowing a significant excess of swimming-associated GI, HCGI, skin and totalillness symptom rates than the latter. The threshold level for staphylococci was ageometric mean density of 1000 per 100 ml. BA beaches with mean staphylococcidensities higher than this limit exhibited significantly higher swimming-associatedillness rates for skin, respiratory and total illness when compared with RUbeaches.
The swimming-associated symptom rates at BA and RU beaches are presented
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Table 6. Determining the threshold E. coli and staphylococci levels fordifferentiating 'relatively unpolluted' (RU) and 'barely acceptable' (BA) beaches
(a) E. coli
RU/BAt
1/82/73/64 / 55/46/37/28/1
Dividing lineE. coli/100 ml
90}130180250260270340840
(b) Staphylococci
RU/BAf
1/82/73/64 / 55/46/37/28/1
Dividing lineStaph/lOOml
300}350650
10001150125014502200
GI
n*
****n
GI
nn*nnnnn
HCGI
*n*
***n
HCGI
nnnnnnnn
Skin
*
n***nn*
Skin
nn*
****
HCGTandskin
***
n**
HCGIandskin
nnn*****
Respir-atory
nnnnnnnn
Respir-atory
nn******
Totalillness
nn*nnn*n
Totalillness
nnn*****
* Significant difference between the swimming-associated symptom rates for RU and BAbeaches at P ^ 005 (t test).
n, Xo significant difference between the swimming-associated symptom rates for RU and BAbeaches at P sj 0-05 (t test).
t Number of RU or BA beaches.} Midway between the highest geometric mean microbial count of the RU beaches and the
lowest count of the BA beaches.
in Table 7. It is apparent the following perceived symptom groups — GI, HCGI,skin, respiratory and total illness - were pollution-related, as significantly higherrates of these symptoms were observed at BA beaches than at RU beaches.
Relationships between microbial indicator densities and swimming-associated healthrisks
The relationships between swimming-associated symptom rates and geometricmean densities of various microbial indicators in beach waters over the wholephase of study were analysed (see Table 8). The mean E. coli densities werecorrelated with swimming-associated HCGI (r = 0-51) or skin (r = 0-55) symptomrates. The highest correlation (r = 0-73) was however found between E. coli andcombined swimming-associated HCGI and skin symptom rates. Faecal coliforms,faecal streptococci and enterococci showed lower correlation with HCGI or skinsymptom rates when compared with E. coli. On the other hand, the geometricmean staphylococci levels were significantly correlated with swimming-associatedear or sore throat symptom rates. When compared with other microbial indicators,
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Fig. 3. Linear relationships between swimming-associated gastroenteritis and skinsymptom rates and geometric mean E. coli densities at nine Hong Kong beaches. Theline of best fit was calculated by the method of least squares. , Regression line:
, 95 % confidence interval.
staphylococci showed the highest correlation with total illness; and were the onlyindicator having a positive correlation with respiratory symptom rates.
A linear relationship could be established between geometric mean E. colidensities and combined swimming-associated HCGI and skin symptom rates forindividual beaches (see Fig. 3). The regression equation was Y = 10-8 log X— 14-89(P ^ 0-05), where X was the mean E. coli density over the whole study period andY the symptom rate.
No significant equation relating E. coli to HCGI or skin symptom rates: orstaphylococci to ear or sore throat symptom rates was found.
DISCUSSIONThe present study shows swimmers at the coastal beaches of Hong Kong are
more likely to complain of GI, HCGI, ear, eye, skin, respiratory and total illnessthan non-swimmers; and swimming in the more polluted beach waters carries asignificantly higher risk of contracting GI, HCGI, skin, respiratory and total illnessthan in cleaner ones. These findings clearly contradict a conclusion of theCommittee on Bathing Beach Contamination of PHLS [4], that 'bathing insewage-polluted seawater carries only a negligible risk to health, even on beachesthat are aesthetically very unsatisfactory'.
Indeed, the study suggests illness associated with swimming is a public healthproblem in Hong Kong. The swimmers in Hong Kong are exposed to a higher riskof developing HCGI symptoms than non-swimmers by 5 times; and for GI. eye.skin and total illness, 2 ^ times. Even with the low sj'inptom rates, the large
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
Beach water pollution in Hong Kong 157number of bathers at Hong Kong beaches means the total incidences of variousperceived symptoms are very high. For Repulse Bay, the most popular beach, ithas been estimated in one bathing season, 14000 people suffer from GI symptomsand 58000 suffer from some form of illness. Over the whole beach-goingpopulation, no fewer than 400000 illness incidences attributable to swimmingoccur each year.
The swimming-associated symptom rates obtained in the Hong Kong study,being the first of its kind in Asia, have been compared with those of similar studiesat USA, Egypt, Canada, Israel and England [16,17,20-24]. For beaches withsimilar degrees of faecal pollution (as reflected by E. coli or enterococci densities),the gastrointestinal symptom rates for Hong Kong beaches (4-1 per 1000swimmers on average) are generally lower than those reported for the marinebeaches of Israel, England and USA. The rates for Hong Kong beaches are moreakin to those obtained for the local bathers at Alexandria beaches. The skinsymptom rates for the marine beaches of Egypt, England and USA are higherthan those for Hong Kong beaches. The total illness rates for the Canadian andHong Kong beaches are similar, namely in the range of 20—30 per 1000 swimmers.
The incubation periods and duration of gastrointestinal symptoms amongst theswimmers in Hong Kong are similar to those observed for the bathers at USbeaches [16,38]. This may suggest swimming-associated gastroenteritis in HongKong and USA is caused by similar aetiologic agents. Cabelli [16,38] hassuggested they are viral in nature, being Norwalk virus and rotavirus originatingfrom human sewage. He has also postulated the higher gastrointestinal illnessrates obtained in the US study than in the Egyptian study could be attributed tothe disparities in the immune state of the populations to these aetiologic agents.The lower incidence of swimming-associated GI symptoms for local bathers thanfor visitors from Cairo at Alexandria beaches would also be due to the higher levelof immunity developed in Alexandria residents to infection by such viruses. Itmay be possible that the low incidence of GI symptoms amongst the swimmers inHong Kong is due to the immunity developed in the local population to entericviruses, because of repeated exposure through various faecal-oral routes sincevery early age.
Children below 10 years of age were found to have significantly higher symptomrates of GI, HCGI, skin, respiratory, fever and total illness than older bathers inHong Kong. This suggests the levels of immunity to the infectious agents causingswimming-associated illness are different between infants and the rest of theswimmer population. Studies at USA, Egypt and Israel have also shown a similarphenomenon, in respect of gastrointestinal symptoms [16,23,38].
The present study indicates E. coli is the best indicator of the health risksassociated with swimming in Hong Kong beaches. This bacterial species showedthe highest correlation with combined swimming-associated gastroenteritis andskin symptom rates when compared with other faecal indicators, namely faecalcoliforms, enterococci and faecal streptococci.
A quantitative relationship between E. coli and the combined symptom rateshas also been established, for setting health-related beach-water quality standards.This together with those indicator-illness relationships reported by overseasworkers are summarized in Table 9. Cabelli [39] has suggested the enterococci level
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
in marine bathing water is the best single measure of its quality in relation to therisk of infectious disease. Table 9 however reveals E. coli is a better indicator interms of correlation with swimming-associated health risks, for the coastalbeaches of Hong Kong and Egypt. Whilst enterococci are the best indicator for theUS marine beaches, this is obviously not a universal phenomenon.
An important point Table 9 also helps to illustrate is that it may not beappropriate for a country to adopt the microbial beach water quality standardsdeveloped by other countries, particularly if they were far away geographically.The immune state of the populations to the aetiologic agents of swimming-associated disease is probably different, and so are the indicator-illnessrelationships. It is important individual countries should conduct their ownepidemiological studies, so that health-related bathing-water quality standardswhich suit their particular conditions could be developed.
The Hong Kong study is the first one which has grouped the swimming-associated symptom rates for HCGI and skin together, and studied their degree ofassociation with microbial indicator densities. This grouping is justified becausefirstly, the HCGI symptom group (or acute gastroenteritis) by itself alreadyconsists of a multiple of symptoms, including gastrointestinal illness and fever.Secondly, there were reports of enterovirus (in particular coxsackievirus)infections amongst bathers, and the syndromes of these viral infections includeamong others, gastroenteritis, respiratory, fever, rashes, and hand-foot-and-mouth disease [40]. The strong correlation between E. coli and combined skin andgastroenteritis symptom rates may suggest enteroviruses, in addition to Xorwalkvirus and rotavirus as suggested by Cabelli [16, 38], are possible aetiologic agentsof swimming-associated illness; and E. coli is a good indicator of such viralinfections among bathers.
The present study points to the need of using more than one microbial indicatorfor denning bathing water quality adequately and setting health-related standardsin Hong Kong. This secondary indicator should not be another faecal indicatorsuch as faecal streptococci or enterococci which relates only to gastroenteritis andskin symptoms, but less well than E. coli in this regard. It should be associatedwith other swimming-associated symptoms. The obvious choice is staphylococci,which showed the best correlation with ear, respiratory and total illness whencompared with other microbial indicators. This indicator is to complement, butnot to replace E. coli, as it cannot be used for predicting swimming-associatedillness rates amongst the bathers in Hong Kong. The adoption of both E. coli andstaphylococci as indicators of water pollution is also justified in that they arepoorly correlated with each other, and may give different pictures on the level ofmicrobial pollution of the same beach (see Tables 1-2).
Geometric mean densities of 180 E. coli per 100 ml and 1000 staphylococci per100 ml are the thresholds for differentiating 'barely acceptable' from 'relativelyunpolluted' beaches in respect of swimming-associated GI, HCGI and skinsymptom rates for E. coli; and skin, respiratory and total illness rates forstaphylococci. These thresholds are recommended for adoption as new microbialwater quality objectives for the bathing beaches of Hong Kong. They shouldbecome guidelines for designing future wastewater treatment and disposal |facilities at or near bathing beaches. 1
L Jhttps://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
The proposed bacterial water quality objective of a geometric mean E. colidensity of 180 per 100 ml corresponds to a combined swimming-associated HCGIand skin symptom rate of 10 per 1000 bathers (see Fig. 3). A geometric mean E.coli density of 610 per 100 ml (equivalent to 60% compliance with the 1000 E. coliper 100 ml criterion), which has already been adopted as the dividing line between'barely acceptable' and 'unacceptable' beaches in Hong Kong, corresponds to aswimming-associated HCGI and skin symptom rate of 15 per 1000 bathers. Thisrepresents the maximum acceptable risk of illness connected with swimming inHong Kong beaches, unknowingly adopted by the Hong Kong community. On theother hand, those beaches with a geometric E. coli density of 24 per 100 ml or less,where the swimming-associated HCGI and skin illness symptoms would beundetectable, can be regarded as having 'good' water quality.
Based on the above observations, a new annual four-tier ranking system hasbeen developed for the coastal beaches of Hong Kong (see Table 10). Rather thanjust designating the beaches as 'acceptable' or 'unacceptable', they are classifiedinto 'good', 'acceptable', 'barely acceptable' and 'unacceptable'. The public arepresented with the information so that they can make the choice themselves as towhich beach they should go for swimming, after taking into account the healthrisk levels involved. It is notable the acceptable risk criterion for the beaches ofHong Kong is 15 gastroenteritis and skin symptoms per 1000 swimmers, whilst thecriterion recommended by US EPA [18] for the coastal beaches of USA is 19gastroenteritis symptoms per 1000 swimmers.
This system for ranking beaches according to health risk levels associated withswimming was first used in Hong Kong in late 1988 [41], and accepted by themunicipal councils, the beach management authorities. The number of bathingbeaches with water quality classified as 'good' was 9; 'acceptable', 19; 'barelyacceptable', 7; and 'unacceptable', 7 in that year.
ACKNOWLEDGMENTSWe thank our colleagues in the Department of Community Medicine of
University of Hong Kong for obtaining the epidemiological data; and in theEnvironmental Protection Department for collecting water samples and under-taking microbiological analysis.
We are also grateful to the Director of Environmental Protection, Hong KongGovernment for permission to publish this paper.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
1. Laws of Hong Kong, Chapter 358. Water Pollution Control Ordinance. Statement of waterquality objectives, Southern, Junk Bay and Port Shelter Water Control Zones. Hong Kong:Government Printer, 1980.
2. World Health Organisation/United Nations Environment Programme. Health criteria andepidemiological studies related to coastal water pollution. Copenhagen: WHO RegionalOffice for Europe, 1977.
3. United Nations Environment Programme/World Health Organisation. Assessment of thepresent state of microbial pollution in the Mediterranean sea and proposed controlmeasures. Geneva: United Nations Environment Programme, 1985.
4. Public Health Laboratory Service. Sewage contamination of coastal bathing waters inEngland and Wales: a bacteriological and epidemiological study. J Hyg 1959; 43: 435-72.
5. Hawley HB, Morin DT, Geraghty ME, Tomkow J, Phillips CA. Coxsackievirus B epidemicat a boys' summer camp: isolation of virus from swimming water. JAMA 1973; 226: 33-6.
6. Denis FA, Blancrouin MD, DeLignieres A, Flamen P. Coxsackie A16 infection from lakewater. JAMA 1974; 228: 1370-1.
7. D'Alessio DJ, Minor TE, Allen CI, Tsiatis AA, Nelson DB. A study of the proportions ofswimmers among well controls and children with enterovirus-like illness shedding or notshedding an enterovirus. Am J Epidemiol 1981; 113: 533^1.
8. Bryan JA, Lehmann JD, Setiady IF, Hatch MH. An outbreak of hepatitis-A associatedwith recreational lake water. Am J Epidemiol 1974; 99: 145—54.
10 Makintubee S, Mallonee J, Istre GR. Shigellosis outbreak associated with swimming. Am JPublic Health 1987; 77: 166-8.
11. Baron RC, Murphy FD, Greenberg HB, et al. Norwalk gastrointestinal illness: an outbreakassociated with swimming in a recreational lake and secondary person-to-persontransmission. Am J Public Health 1982; 115: 163-72.
12. Koopman JS, Eckert EA, Greenberg HB, Strohm BC, Issacson RE, Monto AS. Norwalkvirus enteric illness acquired by swimming exposure. Am J Epidemiol 1982; 115: 173-7.
13. Stevenson AE. Studies of bathing water quality and health. Am J Public Health 1953; 43:529-38.
14. Federal Water Poll Control Adm, National Technical Advisory Committee. Water qualitycriteria. Washington DC: Department of the Interior, 1968: 7.
15. Henderson JM. Enteric disease criteria for recreational waters. J San Engr Div 1968; 94:1253.
16. Cabelli VJ. Health effects criteria for marine recreational waters. US EnvironmentalProtection Agency, 1983. EPA-600/1-80-031.
18. United States Environmental Protection Agency. Ambient water quality criteria forbacteria - 1986. US Environmental Protection Agency, 1986. EPA-440/5-84-002.
19. Dufour AP. Bacterial indicators of recreational water quality. Canadian J Public Health1984; 75: 49-56.
20. Dufour AP. Health effects criteria for fresh recreational waters. US EnvironmentalProtection Agency, 1984. EPA-600/1-84-004.
21. Seyfried PL, Tobin RS, Brown NE, Ness PF. A prospective study of swimming-relatedillness. I. Swimming-associated health risk. Am J Public Health 1985; 75: 1068—70.
22. Seyfried PL, Tobin RS, Brown NE, Ness PF. A prospective study of swimming-relatedillness. II. Morbidity and the microbiological quality of water. Am J Public Health 1985;75: 1071-5.
23. Fattal B, Peleg-Olevsky E, Yoshpe-Purer Y, Shuval HI. The association betweenmorbidity among bathers and microbial quality of seawater. Wat Sci Tech 1986; 18: 59-69.
24. Brown JM, Campbell EA, Rickards AD, Wheeler D. Sewage pollution of bathing water.Lancet 1987; ii: 1208-9.
25. World Health Organisation/United Nations Environment Programme. Correlation betweencoastal water quality and health effects. Copenhagen: WHO Regional Office for Europe,1986.
(i H Y G TOO
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
26. Joint Committee of the Public Health Laboratory Service and the Standing Committee ofAnalysts. Membrane filtration media for the enumeration of coliform organisms andEscherichia coli in water: comparison of Tergitol 7 and lauryl sulphate with Teepol 610.J Hyg 1980; 85: 181-91.
27. Department of the Environment, Department of Health and Social Security, Public HealthLaboratory Service. The bacteriological examination of drinking water supplies 1982.Reports on Public Health and Medical Subjects No. 71, Methods for the Examination ofWaters and Associated Materials. London: Her Majesty's Stationery Office. 1983.
28. Cheung WHS. Escherichia coli: A bacterial indicator of faecal pollution. In : Proceedings ofthe Joint Meeting of the Hong Kong Society of Microbiology and the Guangdong Societyof Microbiology. 2-5 June 1985; 30-43.
30. Kenner BA, Clark HF, Kabler PW. Fecal streptococci. I. Cultivation and enumeration ofstreptococci in surface waters. Appl Microbiol 1961; 9: 15-20.
31. Slanetz LW, Bartley CH. Numbers of enterococci in water, sewage, and faeces determinedby the membrane filter technique with an improved medium. J Bacteriol 1957; 74: 591-5.
32. Klapes NA. Comparison of Vogel-Johnson and Baird-Parker media for membranefiltration recovery of staphylococci in swimming pool water. Appl Environ Microbiol 1983:46: 1318 22.
33. Alico RK, Dragonjac MF. Evaluation of culture media for recovery of Staphylococcus aureusfrom swimming pools. Appl Environ Microbiol 1986; 51: 699-702.
34. Levin MA, Cabelli VJ. Membrane filter technique for enumeration of Pseudomonasaeruginosa. Appl Environ Microbiol 1972; 24: 862-70.
35. DeVicente A, Borrego JJ, Arrabal F, Romero P. Comparative study of selective media forenumeration of Pseudomonas aeruginosa from water by membrane filtration. Appl EnvironMicrobiol 1986; 51: 832-40.
36. Buck JD, Bubucis PM. Membrane filter procedure for enumeration of Candida albicans innatural waters. Appl Environ Microbiol 1978; 35: 237^2.
37. American Public Health Association, American Water Works Association and WaterPollution Control Federation. Standard Methods for the Examination of Water andWastewater, 16th ed. Washington DC: American Public Health Association. 1985; 985-6.
38. Cabelli VJ. Epidemiology of enteric viral infections. In: Goddard M, Butler M. eds. Virusesand wastewater treatment. London: Pergamon Press, 1981: 291-304.
39. Cabelli VJ. Swimming-associated illness and recreational water quality criteria. Wat SciTech 1989; 21: 13-21.
40. White DO, Fenner F. Medical virology. 3rd ed. London: Academic Press, 1986: 451-78.41. Cheung WHS, Chang KCK. Microbiological water quality of bathing beaches in Hong
Kong, 1988. Environmental Protection Department, Hong Kong Government.EPD/TR5/88. Hong Kong: Government Printer, 1989.
https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0950268800047737Downloaded from https://www.cambridge.org/core. IP address: 54.39.106.173, on 06 Jun 2020 at 02:37:16, subject to the Cambridge Core terms of use, available at
